Exploratory Data Analysis (EDA) has been popular in data-driven research. My question is whether EDA can be used as an effective tool to identify a research question? Please provide the links to relevant examples when available. Thank you.
Yes. This is often called a pilot study. See the attachment for more information. Best wishes David Booth
Depends what you mean, but there are cases where it is not. If I have a dataset that my EDA shows that Group A does better than Group B, if I then create a research question: Which group does better?, the question is already answered so subsequent analyses would not be providing more information (and depending on how you conducted your EDA [e.g., if you searched for dozens of possible difference] would not provide strong evidence). But as David Eugene Booth says, you could treat this as a pilot study and go and collect new data.
@Daniel, With all due respect what you have described is not in any way either a pilot study or EDA. Those things have accepted definitions. What you described is just messing around. Please consult either my earlier attachment and/or the many publications of John Tukey et al on EDA. With Best wishes, David Booth
Yes, you can explore and mine data for relationships and research questions, but you don't publish at that stage. Once you have the hypothesis, you have to test the idea in a new dataset. Only after a few replications should a conclusion be made.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India